Slashdot Mirror
3D GUI Project
Qbertino writes: "A guy that calls himself "matrixnan" introduced this project on NANs Blender homepage. It's gonna be a GPLd 3D GUI for Linux using Blender as construction kit. Blender is a professional freeware 3D Animation/Modelling/Applicationkonstruction kit that features Python as Plugin language (Plugins are a big deal in the 3D business). Coding of the Project uses/will use Python, C and C++. Unlike the 3Dsia project it sticks more closely to the 3rd person perpective of the classical Desktop and avoids going to deep into VR and the acompanied problems. It uses NANs reference grade 3D construcion kit and seems to be on its way quite well - and thus will probably see usability quite soon. Also take a look at some serious eye candy - the screens." I'm a little more skeptical about time frames for actually being able to run this thing, but there are lots of interesting ideas to think about.
...is inherently ambiguous. Why ON EARTH would you want your work, your system files and so on represented so unreliably?
Example: Try using 3DS-MAX for heavy-duty 3D polygonal modelling for a few hours. Now imagine that the undo stack is only a single level deep, and that each mistake you make has deleted a random file from your HDD. Now note that this is a package that's evolved through many years of design by a large team at a well-funded software house.
I know this isn't what you were talking about, but check out this and this. *Laugh*
The keyboard isn't much of a problem, but it's tricky to get a useful rendering of 3d space on a flat monitor (Quake doesn't cut it -- we really ought to have something more immersive to be useful) and a mouse just isn't designed to plot anything other than X,Y mappings. You need some sort of way to plot a point in space -- one of those old Virtual Reality Gloves or something -- to workably interace with a 3d environment.
The problem is, as much work as it's going to take to get something like that going -- with new hardware interfaces of some kind and new software technologies to work it out -- it's still an inadequate solution to the underlying problem. The small step from 2d modeling to 3d modeling is barely worth it when you're trying to get to a representation of a, say, 1000d information space -- especially considering all the work it'll take, both on the part of developers and users.
What we really need to do is rethink how we abstract out the complexity of increasingly powerful information systems. How can we represent this data in a way that humans can grapple with? Making a 1000d interface might be theoretically possible, but people can't even handle 4d models, and some of us aren't even that great with 3d or 2d ones :)
Work like this is IMO a dead end. Until something comes along that really rethinks how we model intricately complex dataspaces, we're going to be sitting on the Desktop plateau for a while, at least in terms of actual progress towards system usability.
DO NOT LEAVE IT IS NOT REAL
Although you can do alot of stuff quickly in the corners of the sceen - the simple fact of life is that most of the real work - the important stuff - happens in the middle. This is one this that seems to be ignored time and time again by interface designers. How to you do the work (in the certre of the screen) without having to move back and forth constantly to change tool, which is really painful. What is really nice are context sensitive tools and menus. The best one i've seen in a while is in a cad program called Archicad. Dependeing on which bit of the model the pointer is over, you can access a pop up floating menu, giving access to tools to modify that particulr section. The menu will float around with the mouse movements while you need it, and when you don't it pops down again.Really simple - really powerful. We really need more creative/contextual tools. Not pretty pretty icons or concentrated tools in the corner of the screen.
lemonade was a popular drink and it still is
Actually, there's five locations on the screen that are all equally fast to reach by mouse. Tog has an excellent article on this, I'd recommend you read it.. It's called A Quiz Designed to Give You Fitts, look at question 3 of the Quiz.
Why? Well I guess the answer is because he can, but what I really mean is why put all ths effort into an interface that, although looks 3d, is really just some pretty pictures pasted onto the same old buttons. All it does is add another level to the Mac v Windoze war ("but I can make better icons in windoze!" - "fuck that - i just boot up a c64 when I get the URGE to edit icons") What about thinking about truely new interfaces that allow people to do/create totally new things? To think in different ways, create new and insightful ideas buy linking infomation in previous unthought of ways. That is what GUI design should be about. If you want to edit icons, stick with the c64. For an example try a piece of software called Revamp. It runs ontop of Propellorhead's Rebirth. If you've got a winbox or a mac and some spare time and an liking of electronic music, give it a go. www.revamp.org
lemonade was a popular drink and it still is
2D Desktops
The desktop usually contains a 2D array of a 1D file list... However there is an implied Third Dimension that there is a depth of content under the surface of each icon. A possible 3D model may place a 2D ground map of file icons, each standing upon a 3D Pedistal which presents the file properties, like a dedication plaque, author photo, etc. Atop the pedastal would be the 3D representation of the content, its scale and depth, its motion of activity, etc.
3D Content
3D is more useful for real 3D content. It'd be great to do RAD development of houses just by a Fantasia of powers... a sorcerer's aprentice lifting walls and stairs at the wave of a hand. Actors animated by interperating the "data sock puppet." Quake designers coould go nuts!
3D Hardware
I've been waiting for videophones long before Max Headroom had them everywhere... but even a great show like that overlooked their use as a VR Interface. If you could go anywhere to use a Videophone booth, you could call your computer and communicate with speech and gesture recognition. Also, you could dial pay services for online videogaming using nothing more than the phone, navigating dungeons and swinging swords... physically or virtually. That kind of "non-appliance VR" is common since Myron Kruger's Videoplace.
Likewise, you could out-do Star Trek if we had 3D webcams and 3D projectors... even the keyboard could become a non-appliance solid-state projection, making the Trek sets look like cardboard.
Having recently spend two weeks building an ultimately unsucessful 3D database visualization system, (Java3D rocks!) I can tell you it's a LOT harder than it sounds, a bit like speech recognition.
The main problem is that there is no consistent paradigm for 3D interaction... no equivalent of a desktop metaphor. Users find themselves lost in space. And such systems are hard to interact with properly with a 2D screen and mouse - the missing degree of freedom create a 'modal' system that cannot be intuitive.
But, if you have to do it, here are Orinoco's tips:
1: Make everything about 20% transparent. You can't work with half the environment hidden behind the other half.
2: View control is the key. Don't make the user have to spin and rotate. Let them pick objects of interest, and then move the camera to a good view of it.
3: Don't try to model a complete 3D environment. Instead, make it "2.5D", with the extra dimension used to express an intrisic scalar quantity rather than a spatial one. eg: A 2-D scatterplot, but each point is instead a bar who's height indicates something.
4: Create a 'groundplane'. Stick to stacking things above this.
5: A 'recursive boxes' scheme, with whole new scenes hidden inside pickable objects works well. (A folder metaphor, if you will)
Just consider the most effective 3D application yet - 3D modelling. Even with a perfect 1:1 correspondence between the visual representation and the underlying model, it still takes experts to manipulate the interface.
Frankly, I think 3D interfaces have to wait until we have cheap and available 3D input devices.
Finally, there's a lot of research that has been done (SIGGRAPH, to name a source) that you would be silly to ignore.
Jeremy Lee | Orinoco
I'm sorry, but this just sounds ridiculous to me. User interfaces are pretty badly designed as it is, adding more mouse buttons doesn't solve anything. It just makes it worse.
Some time ago, someone posted a link to a site discussing user interface design, which discussed some of the great ideas and concepts that are simply ignored. For example, what are the 4 locations on the screen to which you can move the mouse very fast? The 4 corners.. Windows makes use of this to some degree, with the close gadget in the top right, and the start button in the bottom left, but this is useless when windows are not maximized.
Perhaps a bigger problem in this scenario is that of the menus within these windows. If they're not maximized, then to get to the menu in each window you have to click in a different location every time. This is very non-intuitive. I personally love the system on the Amiga workbench, where holding down the right mouse button anywhere inside the app would bring down the previously invisible menu at the top. Since it was at the edge of the screen, you could move and select a menu item very quickly. I'd love to see this system implement again, but I haven't yet.
Oops, kind of got a bit off topic there. I guess my point is, people who design user interfaces should really be looking at some of the great useability studies that have been done, and start implementing them before they concentrate on the eye-candy.
And I see they've also developed a special "DWIM" (Do What I Mean) technology, too:Wow! I hope they get a patent on that, before Microsoft steals it from them!
Slashdot is jumping the shark. I'm just driving the boat.